<!-- This model card has been generated automatically according to the information the Trainer had access to. You should probably proofread and complete it, then remove this comment. -->
bert-base-chinese-finetuned-amazon_zh_20000
This model is a fine-tuned version of bert-base-chinese on an unknown dataset. It achieves the following results on the evaluation set:
- Loss: 1.1683
- Accuracy: 0.5224
- F1: 0.5194
Model description
More information needed
Intended uses & limitations
More information needed
Training and evaluation data
More information needed
Training procedure
Training hyperparameters
The following hyperparameters were used during training:
- learning_rate: 2e-05
- train_batch_size: 8
- eval_batch_size: 8
- seed: 42
- optimizer: Adam with betas=(0.9,0.999) and epsilon=1e-08
- lr_scheduler_type: linear
- num_epochs: 2
Training results
| Training Loss | Epoch | Step | Validation Loss | Accuracy | F1 |
|---|---|---|---|---|---|
| 1.2051 | 1.0 | 2500 | 1.1717 | 0.506 | 0.4847 |
| 1.0035 | 2.0 | 5000 | 1.1683 | 0.5224 | 0.5194 |
Framework versions
- Transformers 4.15.0
- Pytorch 1.9.1
- Datasets 1.18.3
- Tokenizers 0.10.3
